Coimmunization with Antigens Shared between Neoplastic and Normal B Cells Suppresses Induction of Strictly Tumor-Specific Cytotoxic T Cells through In Situ Activation of Regulatory T Cells.

CD4+CD25+ regulatory T cells (Treg) recognize autoantigens and inhibit autoreactive immune responses in a cell contact-dependent manner. In cancer-bearing patients, expansion and functional aberrations of Treg may inhibit immune responses against the tumor. The available evidence suggests that such Treg recognize self antigens expressed by the tumor and argues that induction of anti-tumor T cell responses might be more successful if true tumor-specific rather than lineage-restricted or shared antigens are used for active immunotherapy. Indeed, we have observed a preferential recognition of tumor-individual over shared epitopes by vaccination-induced T cells after immunization of B-NHL patients with recombinant lymphoma idiotype (Bertinetti et al., Cancer Res. 2006). To study this phenomenon in an exemplary fashion, we immunized BALB/c mice with dendritic cells loaded with H-2K-restricted peptides of the immunoglobulin of the A20 lymphoma. A J region-derived peptide served as a model for a shared antigen; a heteroclitic peptide from the CDR3 region represented a tumor-specific antigen. Both peptides bind H-2K^d with similar affinity. Compared to a highly immunogenic influenza HA peptide, the CDR3 peptide was similarly efficient in inducing specific cytotoxic T cells as analyzed by tetramer staining, IFNγ release to peptide stimulation, and in vitro and in vivo cytotoxicity assays with CFSE-labelled, peptide-loaded splenocytes. In contrast, no effector cells were detected with any assay after J immunization. After in vitro restimulation with peptide, however, antigen-specific IFNγ-secreting effector populations were demonstrated for each vaccination, suggesting in vivo inhibition, possibly mediated by Treg, rather than total absence of J-specific T cells. No difference in numbers and the TCR repertoire of CD4+CD25+FoxP3+ cells in the draining lymph node could be detected. However, activation of Treg by J immunization was indicated by potent suppression of antigen-specific splenic effectors compared to CDR3-immunized animals, and by a 4fold higher spontaneous proliferation of FoxP3+ cells from the draining lymph node in vitro. In contrast to CDR3-derived Treg, the addition of J-induced Treg to effector cells resulted in a dose-dependent production of IL-10 in mixed cultures, independently of the antigen specificity of the effectors. Finally, coimmunization with HA and J peptides led to inhibition of the proliferation of HA-specific CD8+ effectors in vivo as demonstrated by adoptive transfer and subsequent flow cytometry analysis of CFSE-labelled TCR-transgenic T cells. This inhibition was absent after coimmunization with HA and CDR3 peptides and could be largely abolished by prior in vivo depletion of Treg with an αCD25 antibody. These data demonstrate in a non-transgenic model that coimmunization with shared and individual, strictly MHC I-restricted tumor antigens leads to a potent inhibition of tumor-specific CD8+ T cells through rapid in situ activation of CD4+FoxP3+ Treg elicited by the shared tumor antigen. It is postulated that these Treg recognize MHC II-restricted self antigens presumably derived from non-neoplastic cells as a consequence of an aborted immune response to the shared antigen. These experiments provide direct evidence that active immunotherapy of malignant tumors exclusively with true tumor-specific antigens has a greater chance of success since the presence of shared antigens will prevent tumor-specific immune responses through Treg activation.